This weekly roundup thread is intended for all culture war posts. 'Culture war' is vaguely defined, but it basically means controversial issues that fall along set tribal lines. Arguments over culture war issues generate a lot of heat and little light, and few deeply entrenched people ever change their minds. This thread is for voicing opinions and analyzing the state of the discussion while trying to optimize for light over heat.
Optimistically, we think that engaging with people you disagree with is worth your time, and so is being nice! Pessimistically, there are many dynamics that can lead discussions on Culture War topics to become unproductive. There's a human tendency to divide along tribal lines, praising your ingroup and vilifying your outgroup - and if you think you find it easy to criticize your ingroup, then it may be that your outgroup is not who you think it is. Extremists with opposing positions can feed off each other, highlighting each other's worst points to justify their own angry rhetoric, which becomes in turn a new example of bad behavior for the other side to highlight.
We would like to avoid these negative dynamics. Accordingly, we ask that you do not use this thread for waging the Culture War. Examples of waging the Culture War:
-
Shaming.
-
Attempting to 'build consensus' or enforce ideological conformity.
-
Making sweeping generalizations to vilify a group you dislike.
-
Recruiting for a cause.
-
Posting links that could be summarized as 'Boo outgroup!' Basically, if your content is 'Can you believe what Those People did this week?' then you should either refrain from posting, or do some very patient work to contextualize and/or steel-man the relevant viewpoint.
In general, you should argue to understand, not to win. This thread is not territory to be claimed by one group or another; indeed, the aim is to have many different viewpoints represented here. Thus, we also ask that you follow some guidelines:
-
Speak plainly. Avoid sarcasm and mockery. When disagreeing with someone, state your objections explicitly.
-
Be as precise and charitable as you can. Don't paraphrase unflatteringly.
-
Don't imply that someone said something they did not say, even if you think it follows from what they said.
-
Write like everyone is reading and you want them to be included in the discussion.
On an ad hoc basis, the mods will try to compile a list of the best posts/comments from the previous week, posted in Quality Contribution threads and archived at /r/TheThread. You may nominate a comment for this list by clicking on 'report' at the bottom of the post and typing 'Actually a quality contribution' as the report reason.
Jump in the discussion.
No email address required.
Notes -
Has this group had any discussion regarding AI use in pornography, specifically, 'deepfakes?' Its come out recently that a major up-and-coming twitch streamer 'Atrioc' (who was recently married, and ostensibly very pro-feminist, but while funny these facts are technically irrelevant to the matter at hand) had been viewing porn 'deepfakes' of multiple major female twitch streamers, including possibly his best friend's girlfriend (last part unconfirmed but highly possible). He's come out with an apology and its a whole thing but I'm sure this community is more interested with the moral/ethical questions therein than internet-celebrity drama so I won't bore you with it.
The following are my perspectives on a few of the potential questions regarding 'deepfake' porn, and AI porn in general. I'd love to hear what other people think about these perspectives, because my thoughts are currently very incomplete on the issue.
First and foremost, I have a strong intuitive feeling that it is deeply wrong, perhaps tantamount to some form of sexual harassment/assault (but of course the non-violent sort) to make 'deepfake' pornography of other non-consenting people. For example taking a picture from a celebrity's instagram and using AI to transform it into a high-fidelity (but technically fake) nude picture of them seems functionally the same as i.e. spying on them in the shower or when they're changing, which are actions I think we can all agree would be some form of wrong or illegal sexual violation (or perhaps we can't? you tell me). The way I think about this is by considering that a strong enough AI program would be theoretically capable of using a clothed picture of someone to actually reconstruct the way the exact way they look naked, which would be quite literally equivalent to the aforementioned situation/looking at them with x-ray glasses, etc. which again (I think) we and most people agree would be wrong. And so, less-powerful AI capable of doing something similar seem to be at least on that gradient of wrong, if not exactly as bad.
Furthermore, AI that actually transplants people's faces onto video depictions of sexual intercourse (which is ostensibly what 'Atrioc' was caught doing) seem worse, or maybe just bad in a different way. I don't have a similar thought experiment to justify why I feel that way but the wrongness of it is my strong intuition nonetheless.
However, I can also sort of see the argument, at least abstractly, that it's a victimless crime. On the other extreme of the spectrum, fantasizing in one's own imagination about the way people look when they're naked, or how it might feel to have sex with them, is not only generally recognized as a very benign behavior, but is also known as something almost everyone does, men and women both. Sometimes, people do this even completely unconsciously, i.e. in their dreams. And what's the difference between looking at a very (or fully) realistic recreation of the way someone might look with their clothes off, and using one's own imagination to do so? What if one's own imagination was very vivid, and you had seen many naked people before thus making your training data very good, and so you also could reasonably expect to make a relatively accurate recreation of the way someone looked while naked, only in your own mind's eye?
The thing is, acknowledging these potential similarities between an action I find morally acceptable and the one I find morally wrong, still doesn't make my intuition about the wrongness of 'deepfakes' any weaker. I feel like there must be some thing that I haven't considered about it yet, which is where I'm hoping you guys might have insight. The only distinction I've found somewhat convincing so far is maybe that the mass-distribution via the internet is what makes it wrong? In other words I find it less wrong (but still wrong somewhat) to make a highly/fully realistic nude of someone and keep it entirely on one's own computer, more so than I find it wrong to make such an image and then distribute it online. This is especially weird because the former is even more apt of a comparison to i.e. peeping on someone in the locker room which is obviously (?) wrong. So why does it seem more okay to me? Help!
I have a few potential explanations that I'm considering as candidates for the source of my cognitive dissonance here:
Perhaps in reality none of the aforementioned actions are wrong. It's not wrong to spy on someone in the locker room, and so it's not wrong to use 'x-ray glasses' to see through their clothes, or use an AI to edit a picture to do functionally the same thing.
Perhaps instead, in reality it actually is wrong to imagine or fantasize about what other people look like while naked. The reason this is so commonly accepted as benign is because its so unenforceable to prevent. But if sexual mores are so arbitrary/constructed that something that would otherwise be wrong can just be arbitrarily agreed-upon as acceptable just because its unenforceable, how really wrong can any ('victimless') violation of sexual mores be said to be? And thus how really wrong is the other situation, where one uses AI?
This kind of segues into 3. which is: Perhaps in reality the ultimate causes of this dissonance are that modern-day sexual mores are completely stupid, so deeply incoherent that acceptance of any one of them will necessarily lead to cognitive dissonance when contrasted against some other. Is the solution to the 'deepfake' issue then to try and change our society's sexual morals/ethics into something more internally coherent?
None of these really address why I feel different about 'turning a clothed photo into a nude' and 'transplanting, in a realistic way, a non-consenting individual's face onto an actor in a depiction of sexual intercourse.' I have no concrete ideas as to why the latter feels overall worse, but also in some other (minor) ways not as bad. And the latter situation is what the whole controversy with the streamer is all about AFAIK. Very confused about all this.
What's right here, and why? What should even be done? Should 'deepfakes' be illegal because of these potential moral/ethical concerns? Should the act of making a deepfake be illegal, or just distributing it? (I think if we wanted to, we could make both of these things illegal. We might not be able to enforce preventing anyone from making them considering the AI-cat is out of the bag, but it still might be worthwhile to have its illegality on the books if it really is wrong. In other circles I'm seeing the claims that a ban would be unenforceable (motivated thinking?) but it seems trivially easy to functionally ban at least the distribution of 'deepfake' porn in a way that would almost certainly actually reduce the dissemination of such porn if not completely eliminate it. Just as i.e. child sexual abuse imagery or zoophilia porn.
I also see a lot of people in other circles being prompted by this discussion to argue about the ethics of AI image generation in general. I generally think this is basically stupid. The arguments which claim that AI image generation is tantamount to plagiarism (of the dataset images, I suppose) are all basically worthless as far as I can tell. But people who have bought into this line of thinking are thus now going as far as to say that i.e. photorealistic porn (even that depicting completely synthetic likenesses) that is generated with AI is a sexual violation (of all the nude or semi-nude women in pictures in the dataset I guess?) Either way I am wholly unconvinced by these arguments and think they basically all stem from a bad understanding of how the AI work, so I don't think I'm super interested in discussing this axis of the debate. But I mention it because this community sometimes surprises me so if anyone here has a really strong argument as to why this might make sense that they think I haven't seen before, feel free to mention it.
Among the responses to this post, one thing that I saw several times was that deepfakes do not affect the person they are made of, and so are ethical, or at the very least, there's no case for regulating them. But I think, as was mentioned at least once, that there is a case to be made that they are comparable to libel. That is, they are able to distort the reputations of people in a negative way. This is bad, and I think is something that can be pointed to as a harm to the person in question.
Furthermore, I think that the graphic nature of a deepfake would probably make it have a more substantial and lasting effect on the perception of someone in the eyes of its viewers than would merely verbal allegations, once fabrications of both varieties were learned to be wrong.
I don't think this is a complete answer to what's going on with my moral intuitions here, because I have a similar gut feeling in this case to someone dreaming up, rather than fabricating, illicit scenarios with someone, which means my intuitions are probably not quite the same as what I have written above, and this is probably to some extent a justification of those intuitions, but I still think this is at least a facet of what is going on that is worth considering.
It's also probably worth keeping in mind that a lot of people care a lot about what people think of them as an end-in-itself sort of thing, even aside from tangible effects on their lives. People want to be liked, respected, etc.
Have a disclaimer saying "everything you are about to see is fake". Libel problem solved.
But I'm not sure that it is, given what I said in the second paragraph there, that this is the sort of thing that might lastingly affect how you view someone, even if you know that it is fake. I might be wrong there, but that seems plausible to me, and that would mean that while, sure, maybe it would get rid of whatever legal claims you could make, just saying that it's fake might not entirely work to prevent it from producing the harmful effects.
Humans aren't perfect Bayesian intelligences, and this might be one place where the differences show up, maybe.
That makes sense, but then you can't use existing notions of libel to justify your intuitions.
I think libel is still useful for thinking about it. Not to say that it violates the laws, and so should be illegal already (well, I assume there's no case for it being libel with the current laws, a lawyer would know better), just to say that the same reason that we might think libel laws are good laws might apply here. I was more arguing that it's fundamentally the same sort of thing as libel is, not that it's actually legally libel.
But again, I'm not sure how much of that is me rationalizing.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link